Weather-predicting atmospheric modulation transfer function

Interference caused by bad weather is rather complex. It produces intensity changes in images and videos that can severely impair the performance of outdoor vision systems deployed for many applications including human based transport operations. Subsequent processing algorithms such as segmentation, feature detection, tracking, object recognition as well as stereo correspondence is greatly influenced by the quality of image data. To make outdoor vision systems robust and resilient under varying weather conditions it is necessary to model the degradation effects and develop appropriate methodology to account for these aberrations. This paper, therefore presents an transfer function based approach using atmospheric attenuation model to predict the degradation in the captured images. This model can essentially compute the variations in environmental irradiance and airlight model used for study of atmospheric scattering in the form of a transfer function. This knowledge will finally help to built appropriate image correction and restoration algorithm for application in outdoor all weather vision problem solving.